| struct S0 { |
| int x; |
| uint a; |
| int y; |
| int z; |
| }; |
| struct S1 { |
| int x; |
| S0 a; |
| int y; |
| int z; |
| }; |
| struct S2 { |
| int x; |
| int y; |
| int z; |
| S1 a; |
| }; |
| |
| groupshared S2 wg; |
| |
| void tint_zero_workgroup_memory(uint local_idx) { |
| if ((local_idx < 1u)) { |
| wg.x = 0; |
| wg.y = 0; |
| wg.z = 0; |
| wg.a.x = 0; |
| wg.a.a.x = 0; |
| uint atomic_result = 0u; |
| InterlockedExchange(wg.a.a.a, 0u, atomic_result); |
| wg.a.a.y = 0; |
| wg.a.a.z = 0; |
| wg.a.y = 0; |
| wg.a.z = 0; |
| } |
| GroupMemoryBarrierWithGroupSync(); |
| } |
| |
| struct tint_symbol_1 { |
| uint local_invocation_index : SV_GroupIndex; |
| }; |
| |
| void compute_main_inner(uint local_invocation_index) { |
| tint_zero_workgroup_memory(local_invocation_index); |
| uint atomic_result_1 = 0u; |
| InterlockedExchange(wg.a.a.a, 1u, atomic_result_1); |
| } |
| |
| [numthreads(1, 1, 1)] |
| void compute_main(tint_symbol_1 tint_symbol) { |
| compute_main_inner(tint_symbol.local_invocation_index); |
| return; |
| } |